2. USCLAIDE IBTICI
THIS DOCUMENT IS BEST
QUALITY AVAILABLE. THE COPY
FURNISHED TO DTIC CONTAINED
A SIGNIFICANT NUMBER OF
PAGES WHICH DO NOT
REPRODUCE LEGIBLY.
wmww T w --.
3. NOTCR: Mken goverment or other dravzng, spedi -
fications or other data am used for ny purpse
other than in connection vith a definitely related
governent po IN t operti on, the U. S.
Govoyrcmnt thereby incurs no responsibility, nor any
obligation vbatsoever and the fact that the Govern-
ment my have formllated, furnished, or in any ay
supplied the said drawings, specifications., or other
data is not to be regarded by implication or other-
vise as in any znner licensing the holder or any
other person or corporation, or conveying any rights
or permission to manufacture, use or sell any
patented invention that my in any way be related
thereto.
4. 0
4co
Research Report
Testing for Extrasensory Perception With a Machine
WILUAM R.SMITH
EVERETT F.DAOLE
MAROARET D.HILL D D C
JOHN MOTT-SMITHr~2 2
Best Available COPY TISA "
cArA SCIENCES LAW3R'ATLRY PROJFCT 860
, c CA DC~. , OfSEADCIH LA3A? C81S. OrFICI OF AEROSPACE 15LARCR, UlII'(O iTAT3S AlI FORI:,L.G. HANSCOM FIRD, MAS
5. Abstract
Parapsychology. of which extrasensory perzettion (ESP) is a branch, is nottransqitory. Since reports of apparently significant ESP perforniarces are increas-
ing in number. it becomes mrore important to use the most rigorous of experimental
techniques in testing the ESP hypothesis.
A~~~~~~~deino .cettcn rigorous experimert in E-S? is presentedasna-ceptable model for work in this field. This report cover. the desin of anoiective
test of three modes of ESP. Pretest considerationis and planning, final testing, andresults obtained using the specially designed testing and recordingmcie h
VERIrAC. Finally, a statistical analysis of the results and a discussion of statisti-
cal considerations art, presented.
7. Tetnoxrsnoy1ecpinwt ahn
1.NRDUrI
Exrsnoypreto sotndsusdb amnadsimssi mtoa
tem eas h rsn eerheieneprann oteeitne rnn
Tedun icsing fppoh.sesr Pretonwha ahn
ttreaeo srykn.rcetdin isten deqiscusedeyx ye dsiists inesigt hpteesotiofa
istesset because it u bprecentreea !inpestiabuin ityto the othtene r n-
eif ene oSPcli schld peoeaapastbe nconclusive' Clisrvn aytosnso ollsequny EsP
if mani co ousil becad oernqiil oe mrie nvgr
a otdrite argumern d oithobe ntes and n-eliPyteve s i n hai eteld
varfl in hisos do isc stion Ho weetach.rdneoaprntmnfsa
PerhapsfESP is matuor sontro erusta ite rnt s Certain ly itie offsorun nofoc
th Mtfany ktuind, andear inhiteraur ttm akomexstotetng di rhypohess fo
theclimmdeforli revnds aE la unertainkn of eo land iopre
iseesene fbe ai ol be aZprocss ofo insialc tlc.Onteohe ad
iofl iclamscould er co-nlues~in-sivey srvn. many th ousd ofdlasand wei-
(eeni vedma fours cudbLInsayd. 1-3
8. 2?
desigred experimezds v4elding e-ative resu'ts have not been reported or. at best.
.eCe!fcdi onwv sight mentiom
B3ecause there exist so many conflicting attitudes relaiZg to ESP in genera!.
, the .- ependent and emotionall ujabiased investigator is forced to-conclude that the
ESP controvery itaS resulted in a stalemate - that neither _:de has conclusively
provn itis case to d~fte-
Para-pa"scho!o-r, of which ESP is a branch, is not transitory. The subject is
attracnng the atternion of an increasing number of respected scientists. With re-
poris zfs.gnificant ESP erformances growing in number :tis apparent that the
ti.c"es fast approaching when objective and emotionally unbi sed scientists, using
ri'rs excrimentai techniques, must test the ESP hypotheses in order to make
clear the nature of these phenomena to themselves, and to-the .ublic. Moreover.
until a sizable segment of the orLhodox scientific comnnnit y o-nspiays the integrity
and co=rage to conduct a seri--s of appropriate experi:mnvts. and publishes both posi-
tive and negative results impartially, the ESP controversy will never be resolved.
Wih this conviction in mind, the authors converged on the problem of experi-
merning in ESP in-order to: (1) test the'r own skills in experimental design k ap-
plyirg them to this little known area; (2) develnp the knowledge which could lead to
intel ige eraluatfions of e.istingstudies and opinions in the field; (3) have first-
hand 7P testing experience with subjects in a controlled er-ironment; (4) gain ex-
perience in analyzing =SP type responses in a rigorous satistical fashion; (5) evalu-
ate various data collection, data reduction, and statistical-anarysis techniq-es for
application to ESP data.
his paper is -I intended to ive aid or comfort to either faction in the ESP
contrcversy. but. rAther. it is a straightfox ward report of an ESP experiment, the
design of wAhich is offered as an acceptable model for work in the-field.
2. RESEARCH TEAM
The research team consisted of a psychologist, an electronics, engineer, a
physicist with a background in statistics, and a mathematician. Each of these dis-
ciplines were represe-ted in the e-ablishment of a rigorous experiment in ESP be-
cause Possible weaknesses in applications of each field had been identified in the
survey of the liter3ture.
it was obvious that the psychological aspects of ESP testing must be accounted
for. .Znsidered. and controlled in expert fashion. Thus. the psychologist was part
of the t-am. in order to develop and implement the instrumentation required for
stiuxLbs control and aiutomatic respense recording. an electronics engineer adept
at desiginrg such sy-tems was reired. The statistical background and competence
of the Jrwcst waS necessary because his knowledge relates t4o experinferltal desg-.
A-
9. the sla- 'xnte"'h~)M-SPSI a---"-ysis i4:trrt~ -
ox flxe UeMta, in the-""a '-.s'th-- es- The: = ematic-a- - rec."dre-I for
The w s romtibWe and wo'-ke- together in- :r -z es de-
U.-p~gh expe-iment Mdcsx'n. ana-y~n the t'ata nreotgi leet
the exncr-Irn-t w'rs' ise and crea~tvl- hase eau- =cznner of the tw-.f Dane:
pated of bis own 4eh'ti u-- Consequen-ti. the teami r---- resulted in the .le:o
of more data and thfe exercase of more ini'tiative than zeenl exnected.
3. FIRT CONSIDERATIONS
Asurvey of the literature definitely disclosed tl- zi-h the proponents and op
ponents of ESP had strong and telling arguments to s~t=Drt their v.iews. it was ob-
,viousthen that these argments had to be takzen into e--tderation w.he~n the expert-
ment was designed.
Opponen-ts of ESP. and even those who suppant the -ancept in vzarving degrees.
emph.esize that a repeatable experiment has never t -esigned in this field. This
stand is Justified from a phvsical sciences siandpoiu.4 at--ere renet-itaoa of exnerinments
and consistency of results.. are mandator'. and histo. ca -- have been avh-evable. If
one can acce$t E-SP as a bonafide phenomenon of natur.ther should be no sound
scientific reason why reneatability could not be oan It rnieht be difficult. but
not impossible.
Opponenits of FS? often claim that ecx' -s are- = conducted in a suffxc cntV'
rigid mannier so that all factors can be cont Th.-d tas:st that tight reins must
be kejt en experimental desigm. the m.anne-r -. which stez~li are uresented to the
e-ubiect. the selection and control of sun-ects. and the caring of restonses which
4 come fr-om the experim-ent.
Theexitence of certain clues in the experimetai environment, whether audio.
visual. or other are frequently seen to he nosstbie in rvof the experimental de-
signs that have been previously --stablished. 1
i'e t-se timran da-ta recorders is
also a real weakness- fromn which many, studie.zswifr -7-- possibilities of biasing
the results, either for or against the hypotheses, or of making up fictitious result-,
exist beca-.se of inade-nae data recording controls in na-y of the previous experi-
ints.
From the very outset of this projct tile invest,1gat zs aiteiflteti to overcomle
these and many other deficiencies observzed in the hisoncail approaches to ESRP re-
search. There was no desire on the part. of the investnz-ors to evaluate thle con-
tept of ESP itself. It was recognized that a single expNerment could not p-ossibly
lead to anyv concluisive evidence either for, or against. e existence of F'3l phenomni..
Consequntly, the mnaineffort was fnc..s& upon the isnof an experiment whielh
coude rneted by, nthr researchers.
10. "ce- ea-- H V
experzmms, Ktse.!nlmnmentiuns that ESP' -r can bervcorA-mj _-=--r-. 3
evaluated c-wo r . ronw cor.r $e tain results '-e4-mhmaocna
n~~~~ts ~f *l~sr~rt~sthe use Oa-nmc~' r for ENKS' .'v m.r-Suing
anti anniamie h eer -izter z - 1 =nAiae .c ,'ofwr
- -= .- ble'tz -r-e am'~c work
a.M. Ire '"'-way aj.,rv!Ore. mveectl to and~.. ila. ~ ch's
cxennv orc :noht!e. speccialI purpase machune capable of Dresentlna CU "I
rad:nix n -- tomatiealiv retndhnc ant;son e'nne o cdrs h
tcaratus so tics-.zne andtix~ as called the V;E!IIAC.
4. APPARATUS
The n: rACis an-%elcncande'eectra-meca-- cal system snecifitv-'
.zesintd -t 1- used in conduct-n ESP xermzs and: consists of:
A -ntrol Conszolc (Wig ce
j~~Z -- -ramnCabi.c (icre2
C 'i-a ilecoreinm'---- ' (Tecltvrewrte" and Tape Rt'nerforator.
rpe2)
M -eactionTi.-nc 11r-e Esterline Anzus Recorder. Eteure 2)
A.nbeects Coasol-e 'r 3)
w"-the s 3. :.nTu ~n tuor :naltinc tests. all components of the Cn
tra! - eci- r tn-t-a"led Min e -n 'rh---ie the Subjects Console isq >rated tnaa-
other roont an appreciable etanee fram, the first. The two room-s are,Situated
:n suchn a way that the subjct, hil at is console. iS unable to abtain an'; clues
as to 'Ana:t 'IS O-ceUrrng. 0.r haS o e in the conro-al room.
u-eSyeeent of the V'ERITAC Na stant-stop. push bu-tton onerated Random
Numue (,eaerntor '.-n-znr u throu'gh 9 zz. - z~-rtcd a. ate of .ppeornratelv
2500 per sec. and whin die generator is stopped.' the last number genecrated is dis-
pl'ayed byv a nen indicator mno-unted on the upper rigt tot, panel. of the Cordra!
Console.
The V'F'TAC zs '-set -t easure the abilm-ty of a subject to correct-;r idertify.
in the abseUnce of any clues, a series of randomly- -eneated numbers. These tnm-
hers aire pres-ented as4 psi Stzmui: in selected modecs of operation and the subjecct
interacts wmith the IVERITAC :nrscr:-e ways. Re-sponses of the sEUbject are zzut-
niatiaaztv timed, ronmare.I mt the s inu s~"ed. an-4 recordt-d.
12. u - _E-
Fiue
AwI
A subject isi tested in the following three moes
1. Clai rvovance: A condition requiring the subject to seOlect the number which
he believes hias beenq generated. The subject -%tar-ts and stops the Random Number
Generator at his own discretion. The numbers. so generatted. are presented as
targets in the Conrol Console, but are not shown to the subject or to anyone else.
2. PvecoL=ition: A condition requiring the subject to select beforehand a
number which he believes -wiill be-generated. The subject starts and stops the Ilan-
dom Number Generatotr at his own discretion. rhe numhers, so generated. are
presented in the Control Console, but are not shown to the subject or to anyone else.
3. Gen -~-4Exrasenscry PorcetAion (GESP): A condition requiring the Sub-
ject, acting as a percipient. to select the number which he believes has been gener-
ated. The numbers, so generated, are presented as targ"ets in the Control Console.
but are not shown to the subiect. il6wever. they are seen by one person who acts
as a telepathic agent or sentler. In this mode clairvoyance or telepathy, or both.
may be operatinz-
Ble ause of -lie hi .h degree of autorn.ticity prolidted Z~ mu VERITAC, a com-
plete man-rnacinni- -condtionis obtained in clairvoyance and precognition tests.
This i uxtremcle', auvarlt aaeuS since it effecti-ely removes any, biasing effects
which could be l-rondueed-. by the investigata or husnan data recorder. In GESP
13. tet-a gnt H-truet ici i-:to thle -Xqwilnxi~ta ea"nvincint, n±,t hn_ oniv IAntis-_
a "e to rva n concentra--7 0pnrtile cozierat ta (n;mbrs
I-eence is made to the Bsic Block Dialz inm ni the MEITAC s~'steni (F-re re
11. Til Control S-ection coamponents are shown abo-v the dotted line, and the Sub-
~Sole is shown below. Nnte that *fte Conro' Console consists of three
aus:mdom Numlber Cenerator Ujt; Se!cto -nd coparator Unit: Data Pro-
cessi-g; itx -Note in addition, that thle remotely 'ocated subject's Console is con-
nected to the Curitrol Console by a cable.
DATA SELECOP RANOO"
CONSOLE
Fieure -1
Recall that the sub~ect attenirs to select the number(s) which hie beheves has
been. or will be. generated. By pushing the appropriate buttons on thle lower panel
of his console (Figuire 5), the subject scts the generated number and Is selected
number into the machine. The generation of a number and the selection of it number
constitute one 'trial sequence-. The order of number generation andi selection is
determined by the type of test being conducted. A switch -inthle Control Console
pro-sets the V1ERITAC circuitry so that the systemi will nlot frunction until the proper
trial sequence order has been. completedi.
Immediately after -Atrial sequence has been properly complted, the-VERITAC
goes into -a'lockiedP condition, and remains locked until tile data in thle machine have
been Printed out by the teletvnewrter- Wifle the vrRiTrAC is in-t'he locked Condition,
the subject can neither generate add-itional mliznbers nor change his selection.
11111I11
14. A
Fi 'tare 51
Two to-ziechnnicai coaiters are miounted on the front parL'a of each con-
sole (Cont-.ol and Suabject's). One counter is labeled 'Trials, and the other is labeled
Correct'. The-., k.'untrs provide a vontinauous tabulation of the number of triais
and the numbher of correct selections mand- by the stibject du:ring a test. If itcc(uss
desired to with-hold this information from the subject, 'he counters in his console can
bie disconnected.
Mlounted in the Programn Cabinet is a digital clock which has been modified for
the VERIITAC systemn. rhe time indicated by the clock is recorded on the data
record sheet for each trial sequence. An example of an actual data record sheet
is shown In Fli.,re 6.
The Reaction Trime Recorder is in Esterline Ang,-s single-pen recorder. By
positioning the pen, the reference line is made to fall in the center of the chart.
Directional changes in the current flow through the instrument will cause the pen
to dleflect to the righit or left of the center line.
When connected to the VERITAC, the pen is deflected to the right when the
Random Number Generator is turned on. and itlhen the -enerat' '5 ttrned off. the
pen drops back to-ward the center line. When a number is saelected, the pen def~e..ts
to the left and then returns toward the center line.
17. The recorder chart*moves at the rate of 6 inches per mim te. or 0. 1 inch per
second. The elapsed time between number generation and se:ecton is measured
from the time the needle starts to drop toward the center hne !after the iumber
has been generated!) until it crosses the center line as the number :s selected. A
sample of an actual reaction time recording is shown in Flg--e 7, Measurement
accuracy is exact to 1/2 sec and can be estimated to 1/4 secor4.
5. EXPERIMENTAL DESIGN
As soon as the VERITAC became available, it was checked for operational
effeciiveness and accuracy, and several pilot studies were r-n. These studies
showed that the apparatus would perform as intended, that it was reliable in opera-
tion, and that the generated numbers met rigorous statistical criteria for random-
ness.
The experiment was then designed with the VERITAC as the key element. Hlow-
ever, this was not a restriction because the machine had been designed to fulfill
the basic requirements set forth in the early planning phases of the project.
Three types of psi were to be investigated: Clairvoyance. Precognition, and
General Extrasensory Perception (GESP). For convenience, the test for Clairvoy-
ance was titled Type C; for Precognition, Type P; and for GE-P. Type G. Each
subject was to complete five runs of.each type, and each run was to comprise 100
trials. Therefore, each subject was to make 500 psi responses - i order to complete
a test series of one type, and 1500 psi responses to complete the test series of all
three types.
In 1950. Margaret Knight wrote: "It is a waste of time to conduct further
laborious experiments merely to demonstratc the occurrence of .IP. This has now
been established beyond reasonable doubt. The aim of future experiments should
be to elucidate the conditions of its occurrence, for in this way alone can we hope
to explain its nature. *
We agree with Knight that more than enough has been published which purports
to present evidence for ESP. More important now would be good experimental
evidence to explain the-mechanism which makes ESP work. However, before the
triggering mechanism can be identified, the phenomena must be observed.
If our experiment could uncover several subjects who apparently displayed
ESP capabilities, their personalities might be similar. Certain personality traits.
or profiles of traits, might be good clues in determining to a first approxination
which type of people could best perform in an ESP situation. For the identification
of personality traits the Minnesota Multiphasic Personality Inventory was admin-
istered to al! s-bjects during the experimental period.
1!
AM -
18. A ~ vorsmc tontrn isa:ot.pie env-rmcrx, which ti;.e tests are
A as nz,- 0osd to the_ cold
be a oMn s_ - Z_- b'-. somewhat remote
'M t- i' stn utt".,,_ an ofl~e acUz;, .. 30 that 4istract.na !minflces
-rrnt'n- o- be Z Me!
Location~ of the testir.L rooms so that no itra'c~lon between the Subjects
i)te nf-Vlali:a-s or -he control eCruument would*he possibie
re eciteria for the subjerts compr-isiog the exper mental sample were estab-
led as follows;
Iror a statistical stamlo-t. was desiral'e to :-mv- a sample of at least
_':biects. IPeople who clainm be gifted, or arc retc ~t-obegifted. with psi
are difficult t- flind. Therefore. i became appare itat a sample of appro-
-t" S'Ize would haE-e to be sielected From a population witch by most criteria would
nine ~thoitiesnpaapsy -olo,believe tat the ESP process is often
xt"- are. Additionally, women are sometimes believed to be more intuitive
Therefore, i was felt thit oUr sample responses would be optimized if
*women were tested.
It -;s essential that the subjects be open, friendlY, cooperative volunteers.
hs'bjcts;ntialvwould be highly iotivalt!d to make the experment a success,
tne chi -dore of motivation could probably he sust ined without too much
iUv alsit- --fet this type of subject wouid be most apt to b.. receptive to
813P h-vnotheses=
PRETEST PLANNLNG
ithth! crtrafrtetsig -vironment and the expermna sample final-
Aii was concluded that a group of-collei~e or junior colege girls would be the
-opriate sample for experi mertation. Appropriateness in this case related to
:'abiliy as well as to composition of the sample, and the fact that-the required
Mng enivironment could very probably he found on the school campus.
After considerable searching. E~ndicott Junior CoLiee in lBeverv. Massachusetts,
,etlidas an ideal place for the experimenta trials- When the college of-
.swereapproached with the proposal that the experimen- be conducted there.
Oval and fAloprto were immediately gven.
vrclasses were contacted by the exoerimtneters, Dnaelc Hilll. A brief
~ation of psi phenomena. the reaiwes Wo th vement Qnahoierip-
of he equtipment were dven to the g.irls In each ch s~ Qcsriaiis were hnvted
19. ----..-.---
13
and answered, and when volurteers were requested,the response was most grati-
fying.
A short time later each 'oi.rnteer was asked to meet with Dagle. A depth type
interview was conduczed in an effort to idertify -gt perst-n! zharacreristics or
attitudes which were originall-y h-pothesized to relate to the experiment. Each
person being interviewed was rated on a 10-point scale for:
1. Degree of understanding regarding the nature of the experiment
2. Degree of previous experiences thought to be ESP types (number and mag-
nitude)
3. Degree of religiousness
4. Degree of talkativeness in the interview situation
5. Degree of relaxation in thelinterview situation
6. Degree of interest and enthusiasm in the experiment
7. Degree of flexibility and open-mindedness regarding the experiment
8. Degree of belief in the ESP hypotheses (Sheep-Goat classification).
The 'Sh; 'p-Goat' evaluation involved an estimation by the interviewer of the
degree to which each subject believed in ESP, and had an aggregation of personal
characteristics which appeared to be either conducive or nonconducive to effective
performance in the ESP situation. Tlis rating stems from other research in ESP
tSchmeidler and McConnell, 1958). In the 10-point rating system, those rated from
I to 5 were termed 'Goats', while thoae rated from 6 to 10 were termed 'Sheep'.
The eight characteristics or attitudes evaluated in the imerview were to be
correlated with any differences in performance that were significant in the experi-
ment to come. The interview was a nonfirective, probifig kind and it is estimated
that highly reliable evaluations of the sample were obtained. A tabulation of the
gradings is presented in Table 1.
7. TESTING
Actual testing began on 22 January 1962, and was continued every school day
until 31 May 1962. Subjects were scheduled for tests at the most convenient times
and as class schedules would permit. Subjects rarely failed to keep their appoint-
t.,:nts. and there was very little-free time between test sessions. All tests were
conducted by Dagle and Hill.
The testing rooms on the second floor of a small cotiage were ideal, and the
VERITAC was installed in such a way that no feedback or communication-writh the
subject was possible once the test-run had begun. Figure 8 shows the building in
which the test3 were -un and Figure 9 illustrates the floor plan of testing rooms.
Whle tie test was in progress, the dors to the subject's room and the control
room were tightly closed. During Type C and Type P tests, both experimenters
remained in the c.fice. During Type G tests one experimenter acted as the agent
at the Control Corsole (Figure 0). while the other remained in the office.
--t--
21. Figure 10
lnitfajllr 15 subjects -ere enroli~lfor the exnerjient. From this number.
37 subjects completed the entire series of tests. The attrition rate of only 8 sub-jects indicates --hat the high degree of motivatowa saidtrug'tthe-
tire testing period. Of the 8 drop-outs, 4 were clue to illness. 2 to academic dif-
iculties, and 2 to loss of interest.
S. RESULTS
Each subject retained in the final sample completed 5 runs of 100 trials each
in each experimenta! mode. This resuw-ted in a total of 1.500 psi phenomena re-
sponses for ezcb subject. or 55, 500 psi phenomena responses for the total group.
Scores for each subject and fcr the gro-up were analyzed by appropriately rigorous
statistical methods. Tabulations of the scores are contained in Table 2.
It is conventional to calculate the statistical probabilit!ies of chance factors
alone being responsibie for the scores obtained in ESP experiments. Paraiq~ycvchoio-
Osts. in aeneral h-ave established a probability 'eve! of 0. 01 as the criter-on o's-gifcance, and ia our evaluations we hv edt hsCneto
A szimuam -ofour evaluations is preseete herewith:
22. I. Forthe Overall Exoe-riment
Ilits. For the 1.500 responses of a single subject in the entire experinient.
the expected hit scot t-is 150 since for each trial. there is I chance in 10 of making
a hWt. The binomial probab ity ofobtaini ng 172 or more hits is less than 0. 01.
Two subjects (Nos. 6 and 8) obtained scores or' 173, a deviation of -t23 from the
expected score. It is Interesting to note that one subject (No. 4) obtained. a score
of only 121. a deviation of -29 from the-expected score.
Clii Square Test of Significance for the Group. A Chii Squaretest of signifi-
cance wsapplied to the total scores obtained by the group. A Chi Square value
with 37 degrees of freedom, greater than 73. 2 was required for significance (proba-
bility less thnn0.01). The Chii Square value obtained in this ter.was 33.8296
(p=O. 3751).
Chi Sqruare Test of Significance for Each Subject. Fifteen runs of 100 trials
each were made by each subject. The expected score for each run was 10. A Ci
Square ter.with 15 degrees of freedom was applied to each subject's scores. For
significance, a Chi Square value eqlual to or greater than 30. 578 was required. In
no case was this value obtained.
Wt Totals. For the 55, 500 responses made by the group in the entire e)peri-
memnt. the expected number of hits was 5. 500. The total number of hits obtained by
the group was 5. 567. a deviation of only +67 from the expected score.
It. For Type C Tests
Hits For the 500 responses of --single sUbject in Type C tests, the expected
hit score was 50. The binomial probability of obtaining at least 67 or more hits is
less than 0. 01. No subject obtained a hit score greater than 64.
Hlit Totals. For the 18. 5001 responses of the entire g-roup in Type C tests, the
exuected hit score was 1.850. The total hit score obtained -was 1911. a deviation of
+61 from the expected, score.
Clii Square Test of Significance for the Giu . Chii Square test of significance
with 37 degrees of freedom was applied to the series scores obtained by the group
in Type C tests. A Chi Square value of 73.2 was required for sigrificance (p=0.01).
The Chii Square value obtained in this test was 38. 733 (p=0. 3936).
Clii Square Test of Significance for Each Subject. A Chii Squoare test of signifi-
cance with 5 degrees of freedom was applied to the run scores of each subject. A
Chii Square value of 15. 086 was required for significance. Irn no case wis a value
this great obtain~ed in Type C tests.
Ml. For Type P Testgs
Hit.Fr he 00 esonses of a single subject in Type P tasts. . e exi)ected
hit score -s 50. The binomial probabilty of obtaining A7 or mocre hits is lezss than
0. 01 -No subject obtained ai hit score greater than 64.
23. --
17
flit Totals. For he 18. 506 r,=s.onses of the entire group in Type P tests, the
ex.pected total ht. 35n The tutal ht scorte obtained -as 70. a deviation
of '20 fron' the ex-reted score.
ChithS 7z:reTest of -ifcance fot the ;rouo. A Chi Square test ofsign
cance with 3 degrees of rreedoif -6) apphed to the series scores obtained by the
group in the Type P teSts. A Chi Square val e equal to or greater than 73. 2 was
required for sig"-ficanrce W 0. 01). The Ch,Square value obtained in this test
was 31. 733 (p 0.2.20).
I iSquare Test of Sinificance for 'ach Subiect. A Chi Square test of sigfi-
cance with 5 degrees of freedom was applied to the run scores of each subject. A
Chi Square value of 15. C86 was required for significance (p=0. 01). One subject
(No. 6) had a Chi Square value of 17. 551 (p = 0. 095).
IV. For Trne G Tests
Hits. For the 500 responses of a single subject in Type G tests, the expected
hit score is 50. The binomial probability of obtaining 67 or more hits is less than
0. 01. No subject obtained a hit score greater than 64.
flit Totals. For the 18. 500 responses of the entire group in Ty;pe G tests. the
exoected total hit score was 1.850. The total hit score was 1786, a jeviation of
-64 from the expected score.
Chi Square Test of Significance for the Groun. A Chi Square test of signifficance
with 37 degrees of freedom was applied to the series scores obtained !y the group
in Type G tests. A Chi Sqare value of 73. 2 was required for significance (p 0.01).
The Chi Square value obtained in this test was 52. 622 (p =0. 0432).
Ch, S.-r-are Test of Significance for Each Subject. A CI! Square test of signifi-
cance with 5 degrees of freedom was applied to the run scores of each subject. A
Chi Square value of 15. 085 was required for significance (p'0. 01). One subject
(No. 1) obtained a CM Sjuare value of 15. 333 (p = 0. 01).
V. Chi Sauare Test for Goodness of Fit. -
The individual run scores for each type of test were grouped in classes accord-
ing to their frequency of occurrence (how many times was a score of 5. G. 7. 8...12
and so forth obtained), and a Chi Square test for Goodness of Fit was applied to each
type of test. The Ci Square vaiu,!s and probabilities obtained are as folows:
Type C: Cihi Square 14. 099: p = 0. 2951
Type P: Chi Square = 9. 9095; p = 0. 5394
Type G: C1i1Square i-t. 124; p =
0
.29 3 7
A probability velue of 0. 01 was required for significance.
VI. Analsi.s According een-Goat Classification.
Referring to odr .. o-n-Goa . termination obtained from the death interviews
of the volunteers. fo-nd hat nur sample of 37 consisted of 13 Sheen and 19 Coats.
£i
a U
- I
24. --n an.y:s of thc resu-ts ordained by each group resulted inthe following:
ISHEEP
i8 Subjects
h Totals. For the 9000 responses of the entire group in each type of test,the ex,.ctez total hit score was 900. The total hit score and the deviation from the
expected score for each type is:
Tye C: 938: Dev. --38
Type P: 911: Dev. ;-!I
Type G: 865: Dev. -35
Clhi S:are Testof SiifieancefortheGroup. A Chi Square test of significance
It.18 degrees of freedom was applied to the series scores obtained by the groupin each tevo, of test. A Chi Square value of 34. 805 was required for significance
(p 0. 01). The following were obtained:
Type C: Cli Square x 24. 844; p - 0.1355
Type P: Chi Square z II. 133; p =0.8865
Type 0: Chi Square = 35. 933; p a 0.00865
GOATS
19 Subjects
Ifit Totals. Fort:e 9.050 responses of the emire group in each type of test,
the expected total hit score was 950. The total hit score and the deviation from the
expected score for each type is:
Type C: 973; Dev. +23
Type P: 95. Dev.
Type G: 921; Dev. -29
C SquareTe~of Sinance for Gr 2 . A Chi Square test of significance
with 19 degrees of freedom was applied to the series scores obtained by the groupmneach type of test. A Chi Square value of 36 191 was required for significance
iP 0. 01). The following were obtained:
Type C: Chi Square 13. 889; p 0. 7894Ty:pe P: Ci Square 20.600; p = 0. 3650
Type G: Chi Square = 16. 689; p 0.6104
9. D CUSSION
The origin:l irten of ths section was to i.-erpret the statistics. However.since no aexra-chance effects were shown, it would be futile to do so. Nevertheless,
the experime.t has shown that certain characteristics of nrnbn-listic research maylead te inaccurate interpretations of experiments in decision-making, guessing,
or..n:catIGns. and ESP.
The first possibility for interpretive errors involves the use of the word
"-Lgdfca:l.is converdiona in probabilistic experiments to report as Isignificant
i f
25. thoLse e.n- wc lo c-f .1d tne sm2.- Xlv?;ls:s a: -a:lre-Fettrted iw-.,l
Icr X~iffl~.1f t" :" l' for i:51P resea-rrb The xorl !siz.ir-eCnt'mu.
onetMeans MtV, mit nrobable that nofl-chnce fet are a work. R.t
does nim mean Intchanceis dfner:nprtv or that a non-chance:un-4.-n~n
v-ar;-able, or ant. 'has Leen :soijte!. onccar:Aess reauecrs mhzeven nuae
-a tfinmg whirh is siznificant_ at he . lit levvel as shlowing-: th1at the chances are10
to Ithat a non-chance aetwas Operat-ve Thininference is -nu~~-
roabilist"c Ixeienee eetel 00 tinis. 10 of the aeslts couldz sho-
'signa-fi- ant' results b ha" alc-,One. To carrv tht Iogdc one sten firj-r. :
pears as thourgh 11100 of ,Il. ES exz-erznent's that have ever he---%cornducted have
exhibited !-inHflcancta foweer, this would be expected accord.nng to probablil-t-.
laws.
n example of the diffieuhv of interpreting an experiment.- whose sole descriuxion
iS in terms of acceptance or rejection of the null hvrothesis mnay- be fndfrom oar-
own data. The scores of the most promnising subjects. Sheep. were separated frm
the scores of the least promising subjects. Coats. In the C1SP test for Sheen. a
Significant group v value was found- bae n h ul rolsis: the probabilityv
of, the hit is 1 -10th, wmcka nd hare been exceed br chance wit alliho of
n. 90S.One may be ot -eto scetthi:s as a truy ign.c. est eas
thoeticall Sheen are suposed to score more ,.on-chance than fliats. and-here a2
W110to I oossifl-ty is Cnntae.Wan cut th odds don iranedatihe'-cwe m
1-M to Ito IC to 1 br obserinng that thre, tet were aunlied to both Shee-p and
s9attributed to the fact that thi's is the Sheenpru, ti a oanticularlv einotv !Or
of second guinessing unless one predcted this would hannen amk! had!a mouc:l Ir test-
inZ this hyriehc s Two pnrini raus..4 could halve made er.r 'Z -:u e
Oneisthat the average hit probabi:lity for the group was eithriheorlwrha
t110 but tneIsrbto assilabama o..e ho other is tint some people
werepl'e55in- w: a higher probability in the group. while others with a lower
probWailty, the mean of the grou bing anchanged hutth variance of the dist-
nrmaon being th-us increased- Prior to rnigthe experinin. -Ae had no mreans
dividin up the Shen ;imo D-- dninus Sheen roe nght cal Hem. so ouir aherm t've
1,ypoti'ems was mnat the mean of the whole xroup-would be shife eihr up or down.
Calc-ulation of the baOsed on this assuinjtion givea ne signficac Maue o
C-h- a~sd o-9 05wt h 'hnohssh lcation is that -ore-
sangour- alternativC hynothesis is the a:lmronnate cac. nur results are never-
i thelecss. deae to chance.
Another !ugscal error EnS&Pexeies n nanticlar, isintdced 1w the
lackr of ir in seting statistical criteria for selecting esnsfor rpaaiiy
and for scze promising subftr for czr..uous. intensive research, inths
26. 20
role, the null hypothesis test is used to discriminate between two classes of re.Iult;
one, the majority, which is clearly chance; and the other. only a small perceatage,
which may be due to non-chance factors. This process, if correctly irnplemerted,
is desirable and may be the best technique for ultimately testing ESP phenomena.
Even if it is assumad that ESP is a real, albeit illusive phenomenon. this method.
to be effective, must be used with a full knowledge of the dramatic effects that the
statistical criteria impose upon the selection process. For 'his method to be effec-
tive, the significance level must be carefully chosen in relation to the probability
that a non-chance phenomenon may operate in a given experiment.
For example, suppose that ESP operates in 1/1000th of all such experimens.
Then of those experiments which show Isignificant' results at the . 01 level, 9 out of
10 will be due to chance, and only I in 10 will be due to the extra chance factor. A
model designed on the basis of the ten 'significant' experiments will be weighted 90
percent on the basis of chance results. If a significance level of . 0001 is demanded
for the prototypes, the final model would be weighted only 10 percent on the basis
of chance results and 90 percent on the basis of the non-chance findings. The percent-
ages would be reversed by changing twe demanded level of significance from . 01 to
.0001 if the assumption were made that 1/1000 of the experiments showed true results.
At first the manipulation of the significance level may appear strange and some-
how wrong. However, if a non-chance cause Is operating in a particular experiment,
there is no reason in principle why we cannot demonstrate it at any significanze level
no matter how small. In fact, the ability to do this is at the heart of the probabilistic
interpretation of the experiment. The principal effect of choosing a very small
significance level is that the experimerit must consist of a great many trials so that
the result will be suitably discriminated from a possible chance resitlt. If a signifi-
cance level of, for instance, one millionth Is used, a great many results are as-
cribed to chance, and sufficient trials must be taken to insure that the non-chance
result will be less likely still. Minor irritations may also arise when dealing with
small significance levels; such as. the inadequate ranges existing in prepared
tables. and the need to verify that approximate distribution functions are still ap-
plicable. However, these are details which have nothing to do with the principle of
the matter.
It is apparent, therefore, that this procedure will lead to very long test runs.
Parapsychologists often appear to avoid long test runs on grounds that the ESP aulity
may be impaired. This may or may not be true. but it is nct a statistical argument
against long runs. An experimenter may develop his subjects in any way he chooses
and when lie considers them to be ready may start his trial runs for record. As long
as the run length (short or long) and the number of runs are predeterminedi, any a-
mount of meaningful data can be obtained in this manner.
27. In summarv, Utseems impoortant in ESP and other orobabilistic experimen-ts
to insist that a significance level be qelected that is co:-siderabiy smaller than the
frequency of the hypothesized occurrence of ESP in the experiment. Also, long trials
nd many experiments are required in order to satisfactorily distinguish ESP from
ciiance when it does occur, and many experiments are required to get a significant
sample of signi ficant experiments.
Bly implication, it becomes important to make p--b'-c all outcomes of experi-
ments whether they are 'significant' or 'not significarl' so that the entire sample can
be viewed in truly probabilistic terms. To evaluate a pihenomenon by consider-ng
only the signifizcant experiments is a risky method in probabilistic experiments. By
following these recommendations a repeatable experiment in ESP may be developed
and the extensive correlates which are eagerly sought shoueld be stable in regard to
their signii±eant ESP behaviors.
29. rn r C),:N -r r- m 1 1r- C) 0rm -to -r n~0~i
CI t- r - In -G2 LlI -
Ul L
m o o uji co L-r* co co - NE
VI 2I w 2MVtNm .ON M00 NNmooONOO
0 - - - - --- - - -
to q, w - w- w v- - 0 .0 .0 i- - - m
M~ ~-r- )c - -aC)v- -- - - ---to -nto inC .0 V- -
01 01C
U11
Ii I- -- -
31. 25
Acknowledgments
The authors wish to express their appreciation and indebtedness to P'resident
George 0. Bierkoe. Dean Eleanor Tupper. and Dean Otto R. dauke of Endicott
Junior College. for the many courtesies and assistance extended. and for the services
and physical facilities which were gratuitously and graciously provided.
Appreciation and thanks are also extended to the following student-subjects, with-
out whom there could have been no experiment:
Nancy Anderson. Anita Artstean. Linda iBartol. Susan Bradley. Stephanie Bruett,
Carolyn Cluthe. Carvline Collins. Christy Cotsworth, Bonnie Coughin, Sally Danch.
Harriet Davis, Saundra Dubino. Karen Edge. Sharon Eisler, Judith Frechette, 'Mary
Frontiera, Marilyn Green. Patricia Howie, Janet Kane, Christina Knowles, Virginia
Krawood, Barbara Levine, Danya Lucie, Karin Marcuson. Susan 'McMullen. Caroline
M-undy. Patricia Pharo. Rita Reynolds, Barbara Stillwell.1.indi Ta'sncio, Canz-tance
Thomas, Shirley Wankow. B~onn!.- Warner, Elizabeth Wiggins, Linda Wilson, Priscilla
Wilson. Nancy Zeiler.
32. 27
Glossary
Parapsychology, like other specialized fields, has its own unique vocabulary
and idioms. For those unfamiliar with this language. the following glossary has
been extracted from ParapsycholoZy, The Frontier Science by J. B. Rhine and i. G.
Pratt.
Agent: Sender in tests for telepathy, the person whose mental states are to be
apprehended by the percipient. In GESP tests, the person who looks at the stimulus
object.
Average Score: Average number of hits per run.
Call: The ESP symbol selected by the subiect in trying to identify a target.
Chance Expectation = Mean Chance Expectation: The most likely score if only
chance variation is present.
Clairvoyance: Extrasensory perception of objective events as distinguished
from telepathic perception of the mental state of another person.
Deviation: The amount an observed number of hits or an average score varies
from the mean chance expectation or chance average. A deviation may be a total
(for a series of runs) or average (per run).
Displacement: ESP responses to targets other than those for which the calls
were intended.
ESP: (Extrasensory Perception): Awareness of or response to an external
event or influence- not apprehended by sensory means.
Extrachar.:,o: Not due to charce alone.
GE-P (Gere-al Extrasensory Perceotioij: A technique designed to test the
occurrence of extrasensory perception, permitting either telepathy or clairvoyance
or both to oneratc.
33. a28A
P (Probaiiitv): A mathemnatcal estimate of the expected relative frequency of
a given event if chance alone were operative.
Paransch~Moy: A 1livision of psychologyr dealing with behavoa rproa
effects that are demonstrably nonphysical (that is. which do not fall *Aitlhnthe scope
of physical principles).
Perc i'ent = Subject: The person who makes the callS in an E:P iest.
Preogficn: Cognition of a future event which could not be known through
rational inference.
Psi: A general term to identify personal factors or processes which are non-
physical in nature. It at)proximates the popular use of the word *psychic' and the
technical onc 'pa rapsychlea-L'
Piheoena: Occurrences which result from the operation of psi ... (for
example, ESP phenomena).
Run: A group of trials.
Score: The number of hits made in one run.
Total Score: Total of scores made in a given number of runs.
Series: Several runs or ?xperimental sessions that are grouped in accordance
with a stated principle.
Session: A unit of an ESP experiment comprising all the trials of one test oc-
casion.
Significance: A numerical result is signifct when it equals or surpasses some
criterion of degree of chance improbability. The criterion commonly used in Para-
psychology todayV is a probability value of . 01 or less.
Subject: The person who is experimented upon. In ESP tests, most commonly
the percipient (though also the agent in GESP and telepathy).
Target: In clairvoyance or precognition tests the stimulus object; in telepathy.
the mental state of the agent.
Telepath : Extrasensory perception of the mental activities of another person.
Trial: In ESP tests, a single attempt to identify a stimulus object.
34. Bibliography
liridgeman. P. W.. 'Probability. Logic and ESP'. -1-ine., W. 1956, pp. 15-17.
Burdock. E. I., "A Case of ESP: Critique of IPersonzal Values and ESP Scores' by
Gertrude R. Schmeidier". Journal of Abnormal and Social PsycholonvC~. 49, 1954,
pp. 314-315.
Blurington, Richard Stevens. and May, Donald Curtis. Jr., Handbook of Probahility
and Statistics with Tables. Sandusky: Handbook Publishers. Inc.. 1953,
Dahilstrorn, Grant W.. and Welsh. George Schiager. An MM!Handbook, The Uni -
versity of Minnesota Press, Minneapolis. 1960.
Daiton, G. F., "Operative Factojrb on Spontaneo-is Telepathy". Journal of the Society
for Psychical Research. 38. 1956, pp. 287-319.
Ehrerrvald. Jar. *Presumptively Telepathic lncidezfls During Analysis*. Psychiatry
Quarterly. L4, 1950, pp. 726-743.
Ehrenwald, Jan. "Telepathy and the Child-Parent Relationships". Journal of tie
American Society for Psychical Research. 48, 1954. pp. -3-55.
4 ~Ehrernwald, Jan. New Dimensions of Deep Ar-alysis. Grine and Stratton. New York,
1955.
Ehenwald. Jan. "Telepathy: Concepts. Critra n Cneune'.Pyhar
Quarterly. 30. 1956. pp. 425-444.
-is. I~anr Hobbs. M.; and Mull. H. K.. 'Friendship as a Factor in ESP%.
Journal of General Psycholog . 55. 1956. pp. 28l-283.
Feller. William. An introduction to Probability Theory and its Applications. John
Wiley and Sons Inc., New York. 1957.
Fisher. R. A.. Statistical Methods for Research Workers. 10th Ed.. Hatfner. 'New
York, 1958.
Fiske. G. WV.. and West. D. J..*"Towards Accuratde Prediction from ESP Data~,,
Journal of the Society for Psychical Research. 3. 1957. pp. 157-16-1.
*Tables of the Cumulative Binomial Prob-ability Distribution. Harvar- Vtniv'rsity
Press. Cambridge. Massachusetts. 1955.
35. 30
HeywoodI. Rosai--hd. Seyond the Reach of.1se E. P. il-itton. New Vork. 31
ene..John L.. z-An F-.=!uation of E asnoyParctrpton". Poednso
the Am~erican Ph:josophical Society*. _-6. 9DS2~. 513- 5 A3.
Kitselman. A. L., 'llow to Use a D; Ota' C-:mter as a Means of Stnt-ning the Nature
of Time". Privately "Iistributed. 196'..
KrIghl. Margaret. -Theoret-cal !mplicattarz of Teieparhvy'. s-cience News. 18. 1950.
p. Q-20.
McLeish. John. "R1hine of Reason: A Crtztq.e c! ESP" Modern Q~arterly. 1, 1950,
pp. 342-347.
Meii, Paul Ei.. and Scriven. Michael. --Ccrmoauib4iitY of Science and EiSP". Science,
Iu, 1956. pp. 14-15.
MurphY. Gardner, 'Trends *.nthe Study of Eatraserisory Perception'. A_.jmrican
Psv les. 13. 158. pp. 69-76.
Murplhu. Gardner. Challenge of Psychical Research. Harper. New York. 1961.
Price. George. R.. "Scierce and the Superrnarrar'. Science. 122. 1955, pp. 359-367.
Price. George. R.. "Where :s the Definitive Etxperirnent?". Scence, 123. 1956.
pp. 17-18.
Reiser. Oliver. L., Th~e IntS~gj:tion cf Human Knowledge. Porter Sargent. Boston.
1958.
Rhine. .1.S.. The Reach of the %Mind,William Sloan Assoc.. New York. 1947.
Rhine, J. B., New World of the "Mind. Willi~m Sloan Assoc., New York, 1953.
Rhine, J. B.. "The Experiment Should FAt !he Hypothesis". Science, 123. 1956. pp. 19.
Rhine, J. B.. "Comments on 'Science and the Supernatural' ". Science. 123. 1956,
pp. 11-14.
Rhine. J. B.. and Pratt. J. G.. Parapsychology. Frortier Science of the Miffnd. Charles
C. Thomas. Springfield, 1957.
Rhine, Louisa E.. flidden Channels of the Mind. William Sloan Assoc., New York,
1961.
Robertson. L. C.. "The Logical and Scientific Implications of Precognition. Assuming
this to be Established Statistically from the Work of Card Guessing Subjects".
Jou~rnal of the Society Por Psychici Research. 39. 1957. pp. 134-139.
Schmeidler, Gertrude R.. "Personal Values and EiSP Scores". Journal of Abnormal
and Social Psycholor. 47. 1952. pp. 7574-761.
Schmeidler. flertrude R.. and McConnell. R. A.. EiSP and Personality Characteristics.
Yale Uiversity Press. New Haven. Conn.. 1953.
Scriven. Michael. "New Frontiers of the Brain'. Journal of Parapsychology. 25.
1961. pp. 305-318.
Smith. Kendon. and Cannon. Harry J.. 'A Methodological Refinement in the Study of
ESP znd &NegativeFindings". Science. 120. 1354. pp. 148-149.
Soal. S. G.. "The Scientific Evidence for Extrasensory Perception". Ditsqove=y, 1.0
1549. pp. 373-377.
Soal. S. G.. *'SomeAspects of Extrasensory Fercertion'. Proceedings of the Society
for P~ychicali Research. 49. 1951. pp. 131-153.
S GlS .. and B~ateman. F.,*Modern Expc4.ents in Telepathy. Yale University
Press, New Haven, Conn., 1054.
Soal. S. G.. "On 'Science and the Supernaura'. in,1.196 pp.91.
36. V.. ' 't'nkwnhc , G.. I '.I.i I '. r, f;; ,row :% ra to.-nosior 1 -c roa, 4
ChJA 1 ,r'iln*4 tu%1J. 14tt~v, firr.v9I, lk)#tO'f. I ".56e.
'I~.: Aonk in. '-ip:i'ts *It h -ir~~~:j'u2at o#,fr
"M;,i.~KerkaII, loI. Co. An ligrojz( tinfi to~this 'Vhso,v o- u*''W'si .
1:,J. ,If ihwmr. No. w Yrk, 19!511,
Journal of P'ar "wh1bgog. D~uke Ur.".erity J'romg, M~rham, N. 1L47,
Jo'urnal and Prcc-adinpg of the Socluty for Psychical Jtemearrch, I.,;n 1332?.
Journal of the American Siet fo Paychi al Ucsmearvc Now Yvork. 1I'O7.
CO
37. T I
= -.:~ *. *
- ~~ ~ ~ ~ ~ ~ Z Zi---t - - -2
4 1
Utz~~~-
z=- t. - .'- -
a ro
= -03* cl
.0.,
0
0~,-,- :me-- Y-
;.-V-* . = z-Z - -= l;7
oj WS-- - . -a~d
-' 06:. - - I 53
a 3 -Iv7R m zC
A-u -- =
U5.9 V
